forked from M3-Academy/challenge-algorithms-v2.0
334 lines
6.8 KiB
JavaScript
334 lines
6.8 KiB
JavaScript
|
'use strict';
|
||
|
|
||
|
const stringify = require('./stringify');
|
||
|
|
||
|
/**
|
||
|
* Constants
|
||
|
*/
|
||
|
|
||
|
const {
|
||
|
MAX_LENGTH,
|
||
|
CHAR_BACKSLASH, /* \ */
|
||
|
CHAR_BACKTICK, /* ` */
|
||
|
CHAR_COMMA, /* , */
|
||
|
CHAR_DOT, /* . */
|
||
|
CHAR_LEFT_PARENTHESES, /* ( */
|
||
|
CHAR_RIGHT_PARENTHESES, /* ) */
|
||
|
CHAR_LEFT_CURLY_BRACE, /* { */
|
||
|
CHAR_RIGHT_CURLY_BRACE, /* } */
|
||
|
CHAR_LEFT_SQUARE_BRACKET, /* [ */
|
||
|
CHAR_RIGHT_SQUARE_BRACKET, /* ] */
|
||
|
CHAR_DOUBLE_QUOTE, /* " */
|
||
|
CHAR_SINGLE_QUOTE, /* ' */
|
||
|
CHAR_NO_BREAK_SPACE,
|
||
|
CHAR_ZERO_WIDTH_NOBREAK_SPACE
|
||
|
} = require('./constants');
|
||
|
|
||
|
/**
|
||
|
* parse
|
||
|
*/
|
||
|
|
||
|
const parse = (input, options = {}) => {
|
||
|
if (typeof input !== 'string') {
|
||
|
throw new TypeError('Expected a string');
|
||
|
}
|
||
|
|
||
|
let opts = options || {};
|
||
|
let max = typeof opts.maxLength === 'number' ? Math.min(MAX_LENGTH, opts.maxLength) : MAX_LENGTH;
|
||
|
if (input.length > max) {
|
||
|
throw new SyntaxError(`Input length (${input.length}), exceeds max characters (${max})`);
|
||
|
}
|
||
|
|
||
|
let ast = { type: 'root', input, nodes: [] };
|
||
|
let stack = [ast];
|
||
|
let block = ast;
|
||
|
let prev = ast;
|
||
|
let brackets = 0;
|
||
|
let length = input.length;
|
||
|
let index = 0;
|
||
|
let depth = 0;
|
||
|
let value;
|
||
|
let memo = {};
|
||
|
|
||
|
/**
|
||
|
* Helpers
|
||
|
*/
|
||
|
|
||
|
const advance = () => input[index++];
|
||
|
const push = node => {
|
||
|
if (node.type === 'text' && prev.type === 'dot') {
|
||
|
prev.type = 'text';
|
||
|
}
|
||
|
|
||
|
if (prev && prev.type === 'text' && node.type === 'text') {
|
||
|
prev.value += node.value;
|
||
|
return;
|
||
|
}
|
||
|
|
||
|
block.nodes.push(node);
|
||
|
node.parent = block;
|
||
|
node.prev = prev;
|
||
|
prev = node;
|
||
|
return node;
|
||
|
};
|
||
|
|
||
|
push({ type: 'bos' });
|
||
|
|
||
|
while (index < length) {
|
||
|
block = stack[stack.length - 1];
|
||
|
value = advance();
|
||
|
|
||
|
/**
|
||
|
* Invalid chars
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_ZERO_WIDTH_NOBREAK_SPACE || value === CHAR_NO_BREAK_SPACE) {
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Escaped chars
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_BACKSLASH) {
|
||
|
push({ type: 'text', value: (options.keepEscaping ? value : '') + advance() });
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Right square bracket (literal): ']'
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_RIGHT_SQUARE_BRACKET) {
|
||
|
push({ type: 'text', value: '\\' + value });
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Left square bracket: '['
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_LEFT_SQUARE_BRACKET) {
|
||
|
brackets++;
|
||
|
|
||
|
let closed = true;
|
||
|
let next;
|
||
|
|
||
|
while (index < length && (next = advance())) {
|
||
|
value += next;
|
||
|
|
||
|
if (next === CHAR_LEFT_SQUARE_BRACKET) {
|
||
|
brackets++;
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
if (next === CHAR_BACKSLASH) {
|
||
|
value += advance();
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
if (next === CHAR_RIGHT_SQUARE_BRACKET) {
|
||
|
brackets--;
|
||
|
|
||
|
if (brackets === 0) {
|
||
|
break;
|
||
|
}
|
||
|
}
|
||
|
}
|
||
|
|
||
|
push({ type: 'text', value });
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Parentheses
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_LEFT_PARENTHESES) {
|
||
|
block = push({ type: 'paren', nodes: [] });
|
||
|
stack.push(block);
|
||
|
push({ type: 'text', value });
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
if (value === CHAR_RIGHT_PARENTHESES) {
|
||
|
if (block.type !== 'paren') {
|
||
|
push({ type: 'text', value });
|
||
|
continue;
|
||
|
}
|
||
|
block = stack.pop();
|
||
|
push({ type: 'text', value });
|
||
|
block = stack[stack.length - 1];
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Quotes: '|"|`
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_DOUBLE_QUOTE || value === CHAR_SINGLE_QUOTE || value === CHAR_BACKTICK) {
|
||
|
let open = value;
|
||
|
let next;
|
||
|
|
||
|
if (options.keepQuotes !== true) {
|
||
|
value = '';
|
||
|
}
|
||
|
|
||
|
while (index < length && (next = advance())) {
|
||
|
if (next === CHAR_BACKSLASH) {
|
||
|
value += next + advance();
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
if (next === open) {
|
||
|
if (options.keepQuotes === true) value += next;
|
||
|
break;
|
||
|
}
|
||
|
|
||
|
value += next;
|
||
|
}
|
||
|
|
||
|
push({ type: 'text', value });
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Left curly brace: '{'
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_LEFT_CURLY_BRACE) {
|
||
|
depth++;
|
||
|
|
||
|
let dollar = prev.value && prev.value.slice(-1) === '$' || block.dollar === true;
|
||
|
let brace = {
|
||
|
type: 'brace',
|
||
|
open: true,
|
||
|
close: false,
|
||
|
dollar,
|
||
|
depth,
|
||
|
commas: 0,
|
||
|
ranges: 0,
|
||
|
nodes: []
|
||
|
};
|
||
|
|
||
|
block = push(brace);
|
||
|
stack.push(block);
|
||
|
push({ type: 'open', value });
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Right curly brace: '}'
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_RIGHT_CURLY_BRACE) {
|
||
|
if (block.type !== 'brace') {
|
||
|
push({ type: 'text', value });
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
let type = 'close';
|
||
|
block = stack.pop();
|
||
|
block.close = true;
|
||
|
|
||
|
push({ type, value });
|
||
|
depth--;
|
||
|
|
||
|
block = stack[stack.length - 1];
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Comma: ','
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_COMMA && depth > 0) {
|
||
|
if (block.ranges > 0) {
|
||
|
block.ranges = 0;
|
||
|
let open = block.nodes.shift();
|
||
|
block.nodes = [open, { type: 'text', value: stringify(block) }];
|
||
|
}
|
||
|
|
||
|
push({ type: 'comma', value });
|
||
|
block.commas++;
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Dot: '.'
|
||
|
*/
|
||
|
|
||
|
if (value === CHAR_DOT && depth > 0 && block.commas === 0) {
|
||
|
let siblings = block.nodes;
|
||
|
|
||
|
if (depth === 0 || siblings.length === 0) {
|
||
|
push({ type: 'text', value });
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
if (prev.type === 'dot') {
|
||
|
block.range = [];
|
||
|
prev.value += value;
|
||
|
prev.type = 'range';
|
||
|
|
||
|
if (block.nodes.length !== 3 && block.nodes.length !== 5) {
|
||
|
block.invalid = true;
|
||
|
block.ranges = 0;
|
||
|
prev.type = 'text';
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
block.ranges++;
|
||
|
block.args = [];
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
if (prev.type === 'range') {
|
||
|
siblings.pop();
|
||
|
|
||
|
let before = siblings[siblings.length - 1];
|
||
|
before.value += prev.value + value;
|
||
|
prev = before;
|
||
|
block.ranges--;
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
push({ type: 'dot', value });
|
||
|
continue;
|
||
|
}
|
||
|
|
||
|
/**
|
||
|
* Text
|
||
|
*/
|
||
|
|
||
|
push({ type: 'text', value });
|
||
|
}
|
||
|
|
||
|
// Mark imbalanced braces and brackets as invalid
|
||
|
do {
|
||
|
block = stack.pop();
|
||
|
|
||
|
if (block.type !== 'root') {
|
||
|
block.nodes.forEach(node => {
|
||
|
if (!node.nodes) {
|
||
|
if (node.type === 'open') node.isOpen = true;
|
||
|
if (node.type === 'close') node.isClose = true;
|
||
|
if (!node.nodes) node.type = 'text';
|
||
|
node.invalid = true;
|
||
|
}
|
||
|
});
|
||
|
|
||
|
// get the location of the block on parent.nodes (block's siblings)
|
||
|
let parent = stack[stack.length - 1];
|
||
|
let index = parent.nodes.indexOf(block);
|
||
|
// replace the (invalid) block with it's nodes
|
||
|
parent.nodes.splice(index, 1, ...block.nodes);
|
||
|
}
|
||
|
} while (stack.length > 0);
|
||
|
|
||
|
push({ type: 'eos' });
|
||
|
return ast;
|
||
|
};
|
||
|
|
||
|
module.exports = parse;
|